Domain: AI Economics & Strategy
Encodes token position explicitly, often via sinusoids.
Set of all actions available to the agent.
Combines value estimation (critic) with policy learning (actor).
Continuous cycle of observation, reasoning, action, and feedback.
Organizational uptake of AI technologies.
Ensuring decisions can be explained and traced.
AI-driven buying/selling of financial assets.
Returns above benchmark.
A single attention mechanism within multi-head attention.
Logged record of model inputs, outputs, and decisions.
Dynamic resource allocation.
Updating beliefs about parameters using observed evidence and prior distributions.
Fundamental recursive relationship defining optimal value functions.
Systematic error introduced by simplifying assumptions in a learning algorithm.
A narrow hidden layer forcing compact representations.
Storing results to reduce compute.
Detecting unauthorized model outputs or data leaks.
Prevents attention to future tokens during training/inference.
Scaling law optimizing compute vs data.
Models accessible only via service APIs.
Startup latency for services.
Declining differentiation among models.
A theoretical framework analyzing what classes of functions can be learned, how efficiently, and with what guarantees.
Increasing model capacity via compute.
Techniques to handle longer documents without quadratic cost.
Optimization problems where any local minimum is global.
Assigning AI costs to business units.
Predicting borrower default risk.
Measures divergence between true and predicted probability distributions.
Increasing performance via more data.